Yeah, I don't think that we're saying very different things here, maybe having different intuitions on where and when the effort is applied.

I don't imagine that those examples you provide are particularly unstable or need a ton of changes over time. Like everything on Steam they seem to be built as a module or an API that developers are on the hook for integrating and can be trusted to not change and need additional work from anybody later.

I'm sure there's some back and forth for support and something breaks every now and then, but Valve's approach seems to be to frontload the hell out of polish and then iterate as little as possible once everything is up and running. They really, REALLY don't want an army of devs touching things constantly, they seem to favor finding what works and letting it be.

Hell, even their weird curation systems that were outright bad when they rolled out were like this. They were remarkably reluctant to make fast changes and once they found something that kinda works they haven't touched it much. And that's definitely the most dynamic and evolving part of their entire ecosystem, by far.

If anything a few pieces are looping back around to being legacy garbage. I don't understand how trading cards survived not just their own obvious failure but the entire NFT backlash without being touched at all, and a bunch of their profile stuff is an absolute mess of old, stale UI, at least outside Big Picture.

I'm not even mad about it. It's insane how much of a feature set they've been able to deploy with a relatively small team while also spending a ton of time and effort elsewhere.